I had previously argued that the GSEs should be broken up into 10 or so separate private companies. On Monday, Alan Greenspan spoke favorably of the same idea. Yesterday, Richmond Fed President Jeffery Lacker argued that the GSEs needed to be nationalized, then "demonstrably privatized" by selling off the remains, likely into more than just 2 pieces.
Such a plan is probably the least risky to tax payers. It also would result in a cleaner and more permanent solution than some kind of Treasury-backed equity investment.
How might this work?
First we'd need to make the GSEs sellable, which they currently are not, as evidenced by their share price. To do this, there would need to be meaningful deleveraging, which has to come at tax-payer risk, I'm afraid. I think the solution is to use GNMA program. Treasury announces they are tendering some amount of Fannie/Freddie MBS. The bonds tendered would be exchanged for GNMA bonds of the same coupon. This would ultimately be similar to the MEGA program. It could be stipulated that all tendered loans would have to meet standard FHA guidelines, so like most hybrid-ARM securities wouldn't qualify. But so what? We'd get the leverage of the GSEs down by a certain degree.
Note that the tender would be initially profitable for the Treasury, because of the spread between GNMA securities and FNMA/FHLMC MBS, currently just under 1 point on 6% MBS, and 2 points on 5.5% MBS. Say GNMA offered to tender any eligible FNMA/FHLMC MBS for 1.5% of the principal balance. GNMA could tender say 10% of the total FNMA/FHLMC balance (about $4.3 trillion) and collect $6.5 billion up front.
The Treasury then fully guarantees the principal of all outstanding Fannie Mae and Freddie Mac senior securities. More on this in a moment. The Fed also extends access to the discount window for some extended period of time, like 3-years or so, to the new owners of the former GSE businesses.
Finally, all current mortgage guarantees would be owned pro-rata by the new companies. So if we created 10 new "GSEs" they would each be responsible for 1/10th of any losses that arose. In this way, the old agency MBS securities would still trade generically, as opposed to being passed off to a new guarantor.
So here is what the buyers of the GSE's broken pieces would get:
- Relatively certain liquidity from the discount window
- Reduced balance sheet due to the GNMA tender
- Pro-rata ownership of the GSE's tax loss carry forwards
- Freedom from the GSE's old "public" mission of providing affordable housing
- Freedom to price their guarantee however they wish
- Freedom to participate in any segment of the mortgage market, not just "conforming" mortgages
In exchange, the buyer has to put enough capital into the new company to realize bank-type leverage levels. Exactly what those leverage levels should be is up for debate, but it should be enough to engender some degree of market confidence.
The market as a whole would get...
- More capital put into the GSEs (viewed in aggregate) provided by private investors, which is all but impossible now.
- A limit on tax payer's costs. GNMA would collect a profit on the initial tender of MBS, which would offset losses taken in those pools. Plus the government would get cash from the buyers of the GSE pieces. All this would serve to mitigate losses significantly.
- A long-term solution to the GSE problem. If there are 10 new entities, none would be too big to fail. All would have to maintain bank-like capital levels, rendering the system safer.
- Certainty as to the end game with Fannie and Freddie.
As for common shareholders, they become victims of the GSEs declining capital base. Preferred shareholders and sub debt holders would see their interests split among the new firms. But that would surely result in an increase in the valuation from current levels.
Today both Fannie Mae and Freddie Mac management are meeting with Treasury officials. My take is that Treasury is setting a deadline for them to raise new capital before the Treasury acts. I hope they at least consider something along the lines of what I've proposed here.